View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Elsevier - Publisher Connector Structure, Vol. 8, 1167–1178, November, 2000, 2000 Elsevier Science Ltd. All rights reserved. PII S0969-2126(00)00526-8 Crystal Structure of the Escherichia coli Peptide Sulphoxide Reductase at 1.9 A˚ Resolution

Fre´ de´ rique Teˆ te-Favier,* David Cobessi,* enzymatic, posttranslational modifications may have on func- Sandrine Boschi-Muller,† Saı¨d Azza,† tional properties of proteins [2]. The mechanisms that allow Guy Branlant,† and Andre´ Aubry*‡ the cell to cope with this problem have been investigated. As *Laboratoire de Cristallographie et de Mode´ lisation des a possible alternative to the proteolytic degradation of oxidized Mate´ riaux Mine´ raux et Biologiques proteins in vivo, an termed peptide methionine Groupe Biocristallographie and sulphoxide reductase (PMSR, EC 1.8.4.6) was found to reduce †Laboratoire de Maturation des ARN et Enzymologie methionine sulphoxides back to methionine residues in numer- Mole´ culaire ous organisms. This enzyme was first identified in Escherichia University Henri Poincare´ BP239 coli [3], where it has been named MsrA as the product of the 54506 Vandoeuvre-le` s-Nancy Ce´ dex cloned and sequenced msrA [4]. In vitro, PMSR was France shown to restore enzymatic activities of ␣-1-proteinase inhibi- tor [5, 6] or apolipoprotein A-I [7] that had been oxidized in some of their methionine residues. Furthermore, disruption of Summary the msrA gene in E. coli and yeast cells revealed an increased sensitivity of the mutant cells to [8, 9]. Background: Peptide methionine sulphoxide reductases cat- Oxidative damages are found in various diseases and have alyze the reduction of oxidized methionine residues in proteins. led, for instance, to investigations of PMSR activity in brains They are implicated in the defense of organisms against oxida- of patients suffering from Alzheimer’s disease [10]. However, tive stress and in the regulation of processes involving peptide in other cases opposite effects of these modifications are re- methionine oxidation/reduction. These are found in vealed. Numerous examples showed no alteration of the pro- numerous organisms, from bacteria to mammals and plants. tein function upon methionine oxidation, which suggests that Their primary structure shows no significant similarity to any this amino acid may be involved in an antioxidant defense other known protein. mechanism [11]. Hence, PMSR would be essential to renew the ability of methionine residues to scavenge oxidative spe- Results: The X-ray structure of the peptide methionine cies. For example, PMSR was found to be crucial for the viru- sulphoxide reductase from Escherichia coli was determined at lence of the plant pathogen Erwinia chrysantemi [12]. In this 3A˚ resolution by the multiple wavelength anomalous disper- case, the enzyme may function to protect bacteria from dam- sion method for the selenomethionine-substituted enzyme, ages caused by the active oxygen species produced by the and it was refined to 1.9 A˚ resolution for the native enzyme. plant’s defense mechanisms. Functional activation of proteins The 23 kDa protein is folded into an ␣/␤ roll and contains a by methionine oxidation was also observed [13]. Therefore, large proportion of coils. Among the three cysteine residues PMSR could act as a regulator of processes that involve methi- involved in the catalytic mechanism, Cys-51 is positioned at onine oxidation, such as activation of the plasma membrane the N terminus of an ␣ helix, in a solvent-exposed area com- G-ATPase by calmodulin [14] or modulation of the cellular sig- posed of highly conserved amino acids. The two others, Cys- nal transduction process in potassium channels [15]. Also, an 198 and Cys-206, are located in the C-terminal coil. active PMSR domain has been observed in Neisseria gonor- rhoeae PilB, an enzyme involved in type IV pili production [16]. Conclusions: Sequence alignments show that the overall fold Finally, PMSR has been implicated in fruit ripening since it is of the peptide methionine sulphoxide reductase from E. coli the product of the E4 gene of the tomato plant [17]. is likely to be conserved in many species. The characteristics The catalytic mechanism of this enzyme was investigated. observed in the Cys-51 environment are in agreement with the The results of site-directed mutagenesis and biochemical and expected accessibility of the active site of an enzyme that biophysical studies have been recently published [18–20]. reduces methionine sulphoxides in various proteins. Cys-51 These studies highlighted the role of several conserved amino could be activated by the influence of an ␣ helix dipole. The acids that could be located in the active site [18]. They espe- involvement of the two other cysteine residues in the catalytic cially drew attention to the crucial role of a cysteine residue mechanism requires a movement of the C-terminal coil. Several whose mutation annihilates enzymatic activity [18–20]. This conserved amino acids and water molecules are discussed as conserved amino acid was proposed to be the catalytic residue potential participants in the reaction. that attacks the target methionine sulphoxide. In the bovine and the E. coli enzymes, the next steps were assumed to Introduction involve two other cysteine residues and a regenera- tion system [19, 20]. Aerobic metabolism, essential to life for numerous organisms, In the present paper, we describe the three-dimensional generates highly oxidative by-products such as hydrogen per- structure of the E. coli PMSR (referred to as MsrA). We deter- oxide, superoxide anions, and hydroxyl radicals. These highly mined this structure at 1.9 A˚ resolution. This constitutes the reactive species can attack sensitive amino acids, such as first structure of a member of the PMSR family. Its analysis methionine, which is readily oxidized to methionine sulphoxide reveals a unique fold with a large proportion of coils and high- [1]. Early studies focused on the drastic effects that such non- lights the role of several residues that might be involved in

‡ To whom correspondence should be addressed (e-mail: aubry@lcm3b. Key words: peptide methionine sulphoxide reductase; MsrA; MAD; ␣/␤ roll; u-nancy.fr). catalytic cysteine residue Structure 1168

Table 1. Refinement and Model Statistics Resolution rangea (A˚ ) 30.0–1.9 (1.97–1.90) Number of reflections used for R calculationa 56,730 (3,896) a Number of reflections used for Rfree calculation 6,385 (445) Data cutoff F/␴(F) 0.0 R valuea (%) 19.5 (21.4)

Rfree value (%) 21.8 (24.6) Number of non-hydrogen protein atoms 4,605 Number of sulphate ions 4 Number of water molecules 530 Mean B factor, protein main chain atoms (A˚ 2) 23.6 Mean B factor, protein side chain atoms (A˚ 2) 25.2 Mean B factor, solvent atoms (A˚ 2) 28.5 B factor from the Wilson plot (A˚ 2) 27.6 Ramachandran Plot Residues in most favored regions (%) 90.1 Residues in additionally allowed regions (%) 9.9 Residues in generously allowed regions (%) 0 Residues in disallowed regions (%) 0 Rmsd from Ideal Geometry Bond length (A˚ ) 0.007 Bond angle (Њ) 1.33 Rmsd for Isotropic Thermal Factor Restraints (A˚ 2) Main chain bond 1.1 Main chain angle 1.7 Side chain bond 1.8 Side chain angle 2.6 Rms difference noncrystallographic symmetry, on C␣ atoms (A˚ ) A/B (8–192) 0.33 A/C (8–192) 0.28 B/C (8–192) 0.36 a Values indicated in parentheses correspond to statistics in the outer resolution shell. the catalytic process. The MsrA three-dimensional structure side by three ␣ helices, with the insertions of one ␣ helix and is discussed in relation to the catalytic mechanism proposed two antiparallel ␤ strands (Figure 1). by Lowther et al. for the bovine PMSR [19] and the recent Central to this overall fold is a plait motif, i.e., an antiparallel chemical mechanism proposed by Boschi-Muller et al. for the ␤ sheet of four strands ordered 4-1-3-2, where the strands ␤1 E. coli enzyme [20]. and ␤2 are connected by the ␣ helix ␣1 located at the exterior of the roll and the strands ␤3 and ␤4 are connected by the ␣ Results and Discussion helix ␣2 beside and antiparallel to ␣1. Near the N-terminal end, an ␣ helix ␣1Ј is inserted between ␤2 and ␤3, and two short Quality of the Structure antiparallel ␤ strands, ␤3’ and ␤3’’, are observed in the long The three-dimensional crystal structure of a selenomethionine- loop connecting ␣2to␤4. Next to the plait motif, an ␣ helix, substituted form of MsrA has been determined at 3 A˚ resolution ␣3, parallel to ␣2 gets to a segment composed of two ␤ strands, by the multiple wavelength anomalous dispersion (MAD) ␤5 and ␤6, that are connected via a coil of five amino acids. method. The MsrA model was then refined at 1.9 A˚ resolution. This long chain is stretched and bent so that ␤5 interacts in a The final model (Table 1) includes three molecules (named A, parallel fashion to ␤4 at one end of the sheet, while ␤6 interacts B, and C) in the asymmetric unit, i.e., a total of 590 amino in an antiparallel fashion to ␤2 at the other end. The resulting acids, 530 water molecules, 3 sulphate ions located at the roll shape is reinforced by the strong bending of ␤2 near its surface of molecule A, and 1 at the surface of molecule B, C-terminal part. The whole domain is wrapped by two long at the N-terminal side of the molecules. In each molecule, a coils, the 42 amino acids of the N-terminal part (Ser-1–Met- dimethyl arsenic group is bound to the catalytic cysteine (Cys- 41) and the 29 amino acids of the C-terminal part (Glu-182–Ala-

51). No electron density was observed in the 3Fo-2Fc or Fo-Fc 211). Residues of the N and C termini are located opposite to maps for residues 195-211 of molecule B and residues 1–7 each other with respect to the domain. and 193–211 of molecule C. The R factor is 19.4% for data between 30 and 1.9 A˚ resolution, and the free R factor [21] Crystal Packing is 21.6%. Analysis of the model with PROCHECK [22] and Hexagonal crystals of MsrA contain 60% solvent and three CNS_SOLVE [23] showed good stereochemistry and no residue molecules in the asymmetric unit. The monomers assemble to in disallowed regions of the Ramachandran plot [24]. form a horseshoe-like arrangement, with molecule B sand- wiched between A and C. No n-fold axis relates them, and Overall Structure both the crystallographic and the noncrystallographic contact The MsrA model possesses a single domain described as an surfaces are rather small, which is not unusual for an enzyme ␣/␤ roll, composed of a mixed ␤ sheet flanked on the exterior active as a monomer in vitro. The superimposition of the amino Structure of Methionine Sulphoxide Reductase 1169

Figure 1. Overall Structure of MsrA (a) Stereoview of the MsrA C␣ trace with every 20th residue numbered. The N-terminal tails represented in red and green correspond to the different conformations observed in mol- ecules A and B of the asymmetric unit, re- spectively. (b) Stereoview of the MsrA molecule A sche- matized as a ribbon, with ␣ helices colored in maroon and ␤ strands in green. The mole- cule is represented in the same orientation as in (a). (c) Topology diagram of MsrA, drawn on the basis of the cartoon obtained with TOPS [47]. ␤ strands (green triangles) are labeled ␤1–␤6, and ␣ helices (maroon circles) are labeled ␣1–␣3 in the ␣/␤ roll motif. Additional second- ary-structural elements are labeled with prime signs. Panels (a) and (b) were produced with MOL- SCRIPT [48] and Raster3D [49].

acids 8–192 shows conformational differences that are in the loop structures. Although large, such a proportion of coils was range of the overall coordinate error (Table 1). observed in several protein structures, as shown by the repre- Few direct interactions between molecules are observed. sentation of the secondary structure contents for all available The most remarkable one is the disulphide bridge that links protein structures over the conformational triangle [25]. How- Cys-206 of molecule A with Cys-86 of molecule B. No covalent ever, in the case of the MsrA model, the overall proportions of bond is observed with molecule C. The difference between the secondary-structural elements need to be analyzed more three independent molecules is larger when their N- or C-ter- finely. Indeed, when the 42 amino acids of the N-terminal end minal ends are compared. The electron density of the whole and the 29 amino acids of the C-terminal end are omitted, polypeptide chain of molecule A is visible, whereas the model common values for protein structures are observed: ␣ helices, of molecule B lacks residues 195–211, and the model of mole- ␤ strands, and ␤ coils, respectively, represent 27.9%, 25.0%, cule C only contains residues 8–192. and 47.1% of the 140 amino acids of the MsrA central se- quence. The absence of secondary-structural elements in the Structures of the N- and C-Terminal Ends long N- and C-terminal ends is undoubtedly one of the main The ␣ helices (more than four residues) and the ␤ strands features of this structure. comprise only 18.5% and 16.6%, respectively, of the total num- If the structure observed in the hexagonal-form crystals is ber of amino acids. The remaining residues constitute irregular not influenced by crystal contacts and does reflect the struc- Structure 1170

Figure 2. Stereoviews of Two Portions of the MsrA N- and C-Terminal Ends

The 3Fo-2Fc electron density map calculated at 1.9 A˚ resolution and contoured at 1.2 ␴ is superimposed on the model. The figure was generated with MOLSCRIPT [48]. (a) Residues Ser-1–Pro-11 of the N-terminal end of molecule A. (b) Residues Lys-192–Ile-203 of the C-termi- nal end of molecule A. This region is poorly defined.

ture that exists in solution, the presence of two long terminal minal region. Although it contains no ␣ helices and ␤ strands, coils should have a reason. Analysis of the sequence composi- the structure of this long coil (Ser-1–Met-41) is well defined in tion is in agreement with the observed structure since various the electron density (Figure 2a). It is stabilized by numerous prediction tools mainly predicted coil structures in these two hydrogen bonds and van der Waals contacts. The values of the regions. As an example, the program Jpred (version 2 [26]), thermal displacement parameters confirm this relative stability which defines a consensus of the secondary structure predic- since they barely vary around the average (Figure 3). In A, tions obtained with various algorithms, predicts with a high the N-terminal end interacts with ␣1Ј, while in B it interacts accuracy the correct nature and position of most of the second- with the N-terminal end of ␤5 (Figure 1a). This difference arises ary-structural elements for the whole sequence. The major from the rotation of the main chain around the ␺ angle of Leu-8 contradiction is the prediction of an ␣ helix from residues 180 (␺LEU-8A ϭ 129Њ and ␺LEU-8B ϭϪ176Њ). It reveals flexibility of resi- to 191, while a ␤ strand is observed from residues 179 to 182. dues 1–7. The conformational difference probably results from Inspection of the amino acid sequence reveals five proline the crystal packing, since the Ser-1 and Leu-2 NH groups of residues, at positions 11, 16, 21, 23, and 39, that could account molecule A, respectively, interact with the Ser-103 and Asp- for the absence of secondary-structural elements in the N-ter- 101 side chains of a crystallographic equivalent of C, while the Structure of Methionine Sulphoxide Reductase 1171

Figure 3. Plot of the Average Thermal Dis- placement Parameters of the Residue Main Chain and Side Chain Versus the Amino Acid Number The corresponding secondary structures are schematized by arrows and rectangles corre- sponding to ␤ strands and ␣ helices, respec- tively.

Lys-6 side chain of B is involved in a salt bridge with the Asp- bifunctional formiminotransferase-cyclodeaminase [29]. The 13 side chain of a crystallographic equivalent of molecule A. 325 amino acids of FT are folded in two subdomains. The first The C-terminal end is well defined in molecule A, probably one resembles the MsrA core (Figure 4a), mainly in its plait as the result of the disulphide bridge that involves Cys-206. motif (strands ␤1–␤4, helices ␣1 and ␣2). Beyond ␣3, differences Only amino acids 196–200 (overall mean B value: 54.9 A˚ 2) ex- appear larger since FT has additional strands. The strand ␤6 hibit poor electron density (Figure 2b), and this result suggests partially superimposes onto the MsrA strand ␤5. However, if local disorder in this region. On the contrary, no electron den- both are bent, ␤6 in FT goes toward the second domain, which sity is observed from residue 195 in B or from residue 193 also contains a plait motif, while curvature of the MsrA strand in C. These differences reveal flexibility of at least residues ␤5 directs the polypeptide chain at the other end of the sheet 193–211. The C-terminal coil (Glu-182–Ala-211), observed in toward ␤6. its entirety in molecule A only, is weakly anchored to the re- Similarly, 77% of the 140 amino acids of the MsrA central maining part of the structure. Indeed, only two interactions are core can be superimposed onto one domain (the distal lobe) observed: the Pro-208 carbonyl group is hydrogen bonded to of the formylmethanofuran: tetrahydromethanopterin formyl- the Arg-155 side chain, and the Leu-207 NH group is hydrogen transferase (Ftr) (PDB entry: 1FTR) [30], with a Z-score of 4.4 bonded to the Glu-116 side chain. The fact that the C-terminal and a rms value of 3.9 A˚ . Again, the plait motifs of both struc- coil in molecule A is poorly stabilized by the molecule itself tures are similar (Figure 4b), but the differences increase start- and that a large number of glycine residues [19] are found at ing from the second half of strand ␤5, which leads to the second positions 195, 199, 201, 202, and 204 reinforces the hypothesis domain of Ftr (the proximal lobe). of the C-terminal flexibility. This prompts us to correlate the These comparisons show that the topology of one part of requirements of the catalytic mechanism with this structural the tertiary structure was already observed in other proteins, property of the C-terminal end, as discussed in a later section. and they also highlight two distinctive features of the MsrA structure: the large content of coil structures at both ends of Structural Comparisons the polypeptidic chain, discussed in the previous section, and ␤ ␤ Searches for similarly folded proteins have been performed on the spatial arrangement of the 5-loop- 6 chain interacting ␤ ␤ the Brookhaven Protein Data Bank (PDB) [27] by using CE [28] with 4 and 1. For that reason, we conclude that the overall and the three criteria recommended by the PDB: the Z-score architecture of MsrA is unique. [28], the percentage of superimposed amino acids, and the root-mean-square (rms) difference of superimposition. None The PMSR Family of the MsrA closest folds listed by CE were found to fulfill the At the present time, eight organisms have been shown to conditions on these three values to define a similar structure. possess an enzyme that displays a PMSR activity in vitro: However, some results came close and have been carefully Escherichia coli [4], Bos taurus [31], Brassica napus [32], Strep- examined on graphics systems. In particular, when the 42 and tococcus pneumoniae [16], Neisseria gonorrhoeae [16], Sac- 29 amino acids of the long N- and C-terminal coils of MsrA charomyces cerevisiae [9], Erwinia chrysantemi [12], and Homo are omitted, 89% of the 140 remaining residues were superim- sapiens [33]. Pairwise comparisons between the sequences posed, with a Z-score of 4.4 and a rms value of 3.8 A˚ to the of these enzymes showed their high similarity; the minimum formiminotransferase (FT) domain (PDB entry: 1QD1) of the calculated value is 48.8% (30.8% identity) between E. coli and Structure 1172

Figure 4. Stereoviews of the Structural Su- perimpositions of MsrA and Its Closest Folds Calculations were made with CE [28]. The C␣ trace of MsrA residues 43–182 is displayed in red, while the whole C␣ trace of the (a) formiminotransferase domain (PDB entry, 1QD1; [29]) of the bifunctional formimino- transferase-cyclodeaminase and the (b) for- mylmethanofuran:tetrahydromethanopterin formyltransferase (PDB entry, 1FTR; [30]) are colored in green. This drawing was generated with MOLSCRIPT [48].

N. gonorrhoeae PMSRs. The multiple alignment of the eight (62% similarity) with the tomato enzyme and 46% identity (63% PMSR sequences confirms the sequence conservation across similarity) with the sequence from Methanobacterium thermo- species (Figure 5). Only a few small gaps are noticed, at posi- autotrophicum. The sequence from melanogaster tions that correspond to shorter/longer loops or helices in the is the most distant to the E. coli sequence (30% identity, 45% MsrA three-dimensional structure. Furthermore, about 60 similarity). The alignment of the eight sequences of active amino acids exhibit overall similarities throughout the eight PMSRs with the 33 sequences of putative PMSRs reveals that sequences, among which 26 residues are strictly conserved. their various number of amino acids mainly correspond to vari- Altogether, these observations suggest that the PMSRs whose ous lengths of the N- and C-terminal ends. The longest se- activity was characterized may have a similar overall fold. quences belong to multidomain enzymes, as for example the In addition to the 8 above-mentioned PMSRs, 33 putative 58 kDa gonococcal PilB involved in regulation of the pilin tran- sequences were extracted from sequence databases on the scription [34], which possesses an N-terminal thioredoxin-like basis of their sequence similarity to MsrA. These sequences domain and a C-terminal domain of unknown function. On the are found in different organisms in which no PMSR activity is contrary, when the focus is on PMSR central cores, sequences yet reported. We refer to them as putative PMSRs. The primary can be aligned with only a few small gaps. Again, their positions structure remains remarkably conserved across species and are compatible with the overall MsrA fold. even among the three domains: bacteria, archaea, and eu- Among the 60 above-mentioned amino acids that are similar karya. For example, the E. coli enzyme displays 44% identity throughout the sequences of active PMSRs, incorporation of Structure of Methionine Sulphoxide Reductase 1173

Figure 5. PMSR Sequence Alignment The sequences of eight active PMSRs from Escerichia coli (abbreviated: ecoli), Brassica napus (brana), Bos taurus (bosta), Erwinia chrysantemi (erwch), Homo sapiens (homsa), Neisseria gonorrhoeae (neigo), Saccharo- myces cerevisiae (yeast), and Streptococ- cus pneumoniae (strpn) were aligned with PILEUP from the Wisconsin Package Version 9.0, Genetics Computer Group (GCG), Madi- son, WI. The sequences from Neisseria go- norrhoeae and Streptococcus pneumoniae were restricted to the PMSR domains. The residues highlighted by the red boxes are strictly conserved in the selected sequences. Green triangles indicate the residues con- served through the 41 sequences of active and putative PMSRs. ␣ helices and ␤ strands in MsrA are indicated by spirals and arrows, and are labeled as in Figure 1c. The figure was produced with ESPript [50].

the putative PMSR sequences in the alignment brings out 11 dimethyl arsenic group induces local conformational changes strictly conserved residues (Figure 5) and about 25 residues in the active site. Similarly, the MsrA active site could dis- that are either conserved with, at most, four exceptions or play slight differences whether Cys-51 is cacodylated or not. substituted by similar amino acids. A few buried amino acids Searches for other crystallization conditions are in progress. have a conserved aliphatic or aromatic character since they A tetragonal crystal form is currently under investigation [36]. belong to the double layer of hydrophobic side chains usually PMSRs can reduce oxidized methionine residues from pro- found at the interface between ␤ sheets and ␣ helices. Except teins of different sizes. The catalytic cysteine Cys-51 is thus for them, most of the mainly conserved residues delineate a expected to be located either at the surface of the protein, in solvent-exposed concave surface that is thought to be the a large accessible cavity, or in a zone whose flexibility provides active site. This region, includes amino acids that belong to Cys-51 accessibility to bulky partners. Analysis of the three- two highly conserved sequences, residues 48–54 and 182–189, dimensional structure confirms this assumption; the active site described in the next section. These are the footprints of is located in a solvent-exposed region of the protein. It is PMSRs. shaped like a largely opened basin whose surface is mainly covered with aromatic amino acids such as Trp-53, Tyr-82, The Active Site and the PMSR Footprints Tyr-134, and Tyr-189 (Figure 6). In the sequence alignment, PMSR sequence comparisons, site-directed mutagenesis, and these side chains are either conserved or replaced by equiva- biochemical studies on several organisms revealed the con- lent aromatic side chains. However, their presence in the vicin- served cysteine residue (Cys-51 in MsrA) to be the catalytic ity of the catalytic residue that lies at the bottom of the basin amino acid. However, it does not have a free thiol function in does not prevent access to the Cys-51 sulphydryl group. The the MsrA hexagonal crystals. In the three molecules of the solvent accessibility of the sulphur atom when calculated as asymmetric unit, a strong electron density was observed at if the dimethyl arsenic group were absent is about 20 A˚ 2, and the end of the Cys-51 side chain in both the 3Fo-2Fc and Fo-Fc even slight differences in the side chain orientation for a non- maps. This electron density was not interpreted after protein cacodylated cysteine residue should give rather large values. building. It clearly corresponded to a bulky group with an elec- Cys-51 belongs to the first highly conserved sequence that tron-rich atom covalently bound to the Cys-51 sulphur atom. is characteristic of PMSRs, 50GCFWG54. The three central amino Furthermore, searches for the selenium sites in selenomethio- acids, Cys-51, Phe-52, and Trp-53, are strictly conserved even nine-substituted MsrA crystals revealed that the highest peak when the putative PMSR-sequences are considered. Taking found after the 18 selenium atoms corresponded to this un- all the sequences into account, the presence of a minimal known atom, thus displaying anomalous dispersion properties. footprint (A,G)xgCFWg can be proposed from amino acids 48 The group was then identified as dimethyl arsenic. Its presence to 54 in MsrA, where “x” designates a nonconserved residue. results from the crystallization medium, which contained so- This first signature is located in the loop connecting ␤1 and dium cacodylate buffer and DTT. These two species react ␣1 and spreads to the first turn of ␣1. The first residue is an together with cysteine residues to form a (dimethyl arsenic)- alanine or a glycine residue, i.e., a short or an absent side cysteine, as was also observed in the catalytic domain of HIV-1 chain. One amino acid separates it from the next glycine resi- integrase [35]. In this case, the comparison of the cacodylated due, which is conserved except in the putative Caenorhabditis and noncacodylated structures shows that the presence of the elegans sequence, where it is replaced by a glutamine residue. Structure 1174

Figure 6. Stereoview of the Active Site Re- gion of MsrA The figure was produced with MOLSCRIPT [48] and Raster3D [49]. Side chains of the amino acids around the catalytic residue Cys- 51 are displayed as ball-and-stick represen- tations.

Then the strictly conserved sequence CFW follows. These 183 side chain, while the Tyr-82 aromatic ring is stacked with amino acids are also located in the active site. The last glycine the His-186 side chain. The strictly conserved character of both residue of the first footprint is replaced by either a cysteine or amino acids argues for the importance of these interactions. a serine residue in four sequences among the 41 considered. Indeed, Glu-183 and His-186 belong to the second highly con- The side chains of three amino acids, Tyr-82, Glu-94, and served sequence of PMSRs. A consensus sequence was deter- Tyr-134, are positioned on one side of the catalytic residue mined for this region, with the putative PMSRs taken into ac- Cys-51 (Figure 7). The Tyr-82 side chain in molecule B is slightly count. AExxHQxφ, corresponding to positions 182–189 in E. displaced by 0.8 A˚ for the hydroxyl group as compared to the coli, can then be proposed as the second footprint of the PMSR situation in molecules A and C. This can be explained by the sequences, where “φ” denotes an aromatic amino acid. This proximity of Cys-86, which is free in A and C, while in B it is signature describes the entrance in the C-terminal coil after ␤6. involved in an intermolecular disulphide bridge with Cys-206 As already mentioned, Glu-183 and His-186 are tightly linked to of molecule A. In molecules A and C, Tyr-82, Glu-94, and Tyr- the positioning of Tyr-82. The Gln-187 side chain is buried from 134 bind one water molecule. In molecule B, where the orienta- the surface; its N⑀2 group forms a hydrogen bond with the Ala- tion of the Tyr-82 side chain is slightly different, a peak of 182 carbonyl group. The last amino acid of this second footprint positive electron density is observed at a similar position, but (Tyr-189 in E. coli) is either a phenylalanine or a tyrosine residue. the probable corresponding water molecule was not included Its conserved aromatic character is obvious since it is stacked in the refinement since the distance with the Tyr-82 hydroxyl with the strictly conserved Trp-53 of the first footprint. group would be too short for a hydrogen bond (2.0 A˚ ). Tyr-82, Asp-129 is observed on the other side of the Cys-51 side Glu-94, and Tyr-134 are conserved in the 41 sequences. Glu- chain relative to the previously mentioned residues (Figure 7). 94 is the first residue of ␤3. Tyr-134 is located at the end of Its O␦2 atom is situated 4.8 A˚ apart from the Cys-51 sulphydryl the long loop connecting ␣2 and ␤4. Tyr-82 belongs to ␣1Ј, group. In the three molecules of the asymmetric unit, the Asp- which is inserted in the loop connecting ␤2 and ␤3. The posi- 129 side chain binds a water molecule that is hydrogen bonded tioning of this amino acid seems to take place by two interac- with the NH group of Cys-51. Conserved in the sequences of tions: the Tyr-82 NH group is hydrogen bonded with the Glu- active PMSRs, this aspartate amino acid is absent from the

Figure 7. Stereoview of the MsrA Active Site Residues are displayed as ball-and-stick rep- resentations, and hydrogen bonds are indi- cated by dotted lines. The strictly conserved Tyr-82, Glu-94, and Tyr-134 in the vicinity of Cys-51 tightly maintain a water molecule. Asp-129, present in all the sequences of ac- tive PMSRs, binds another water molecule.

The 3Fo-2Fc electron density map calculated at 1.9 A˚ resolution and contoured at 1.2 ␴ is superimposed on the model. The figure was generated with MOLSCRIPT [48]. Structure of Methionine Sulphoxide Reductase 1175

Figure 8. The Catalytic Mechanism of the E. coli PMSR As proposed by Boschi-Muller et al. [20], the Cys-51 nucleophilic attack on the sulphur

atom of the methionine sulphoxide (IA) leads, via rearrangement, to the formation of a sul- phenic acid intermediate and the release of

the reduced methionine (IB). An acidic cataly- sis allows the hydroxyl group of the sulphenic acid to be released as a water molecule, with the concomitant attack of Cys-198 onto Cys-

51 (IIA), followed by the attack of Cys-206 onto

Cys-198 (IIB). Return of the enzyme to its re- duced state (III) is ensured by the thioredoxin (TR) and the thioredoxin reductase (TRR) sys- tem with its cofactor NADPH.

putative C. elegans and Drosophila melanogaster PMSRs since acidic catalysis to allow the release of the water molecule and the loop connecting ␣2to␤4 is shorter. the concomitant attack of Cys-198 (see Figure 8, step IIA). Inspection of the three-dimensional MsrA structure reveals the Structure and Function Relationship presence of two amino acids in the vicinity of Cys-51 that The first step of the catalytic mechanism proposed for the are possible proton donors: Glu-94 and Asp-129 (Figure 7). bovine PMSR is the nucleophilic attack of the catalytic cysteine Furthermore, with Tyr-82 and Tyr-134, these amino acids bind residue on the target methionine sulphoxide, and this leads to two water molecules. One of them might correspond to the the formation of a trigonal-bipyramidal intermediate between position of the water molecule released during the catalytic Cys-72 (equivalent to Cys-51 in E. coli) and the methionine process, while the other one could have a structural role. sulphoxide [19]. A nucleophilic attack by Cys-218 (equivalent MsrA possesses four cysteine residues: Cys-51, Cys-86, to Cys-198 in E. coli MsrA) on the intermediate would then Cys-198, and Cys-296 (Figure 9). The existence of a disulphide release one water molecule and the reduced methionine resi- bridge between Cys-86 of molecule B and Cys-206 in A and due and form a disulphide bond between Cys-72 and Cys-218. the presence of the dimethyl arsenic group bound to Cys-51 Return of the active site to a reduced state would be achieved illustrate the high accessibility and reactivity of at least three of these cysteine residues. However, among these three cysteine by thiol exchange involving the formation of a disulphide bond residues, only those corresponding to Cys-51 and Cys-206 are between Cys-218 and Cys-227 (equivalent to Cys-206 in E. coli proposed to be involved in the catalytic mechanism, together MsrA). with Cys-198 here observed as a free thiol. Cys-198 and Cys- Recent studies of the E. coli MsrA catalytic mechanism [20] 206 are located in the C-terminal coil, which is visible only in are not in accordance with this mechanism and showed that the molecule A. The Cys-198 C␣ atom is found 11 A˚ away from Cys-51 attack on the sulphur atom of the sulphoxide substrate the Cys-51 C␣ atom. This distance is by far too large for a leads, via a rearrangement, to the formation of a sulphenic acid disulphide bridge to occur. Furthermore, the side chain of Cys- on Cys-51 and to the concomitant release of the methionine 198 is directed away from the active site. If the C-terminal end residue (Figure 8). The second step of the mechanism consists conformation observed in the crystal is not influenced by the of the reduction of this sulphenic acid intermediate and leads disulphide bridge involving Cys-206, a conformational change to the release of one water molecule via a double displacement that brings the cysteine residues closer together should be mechanism. It first involves the formation of a disulphide bond envisaged. Flexibility of the C-terminal coil has been estab- between Cys-51 and Cys-198, and this is followed by the forma- lished (see section “Structures of the N- and C-Terminal Ends”). tion of a disulphide bond between Cys-198 and Cys-206. For- Analysis of the thermal displacement parameters along the mation of this bond liberates Cys-51. Finally, return to a fully C-terminal coil concurs with the ability of the Cys-198 region reduced state necessitates a thioredoxin-regenerating system to undergo a conformational change needed for the attack on in vivo. the Cys-51 sulphur atom; main chain B factors (Figure 3) slowly The first step involves the nucleophilic attack of Cys-51 on increase from a value of 19 A˚ 2 for Lys-192 to a maximum value the sulphur atom of the methionine sulphoxide. Prior to this of 60 A˚ 2 for Cys-198 and then decrease to a minimum of 17 A˚ 2 step, Cys-51 is likely to become activated. The presence of for Val-205, with standard values being observed up to the C the thiolate ion should be favored by the environment of Cys- terminus. 51 in the structure since the pKa value of a cysteine residue in The next step of the catalytic mechanism involves the forma- solution is usually too high to allow deprotonation at physiologi- tion of a disulphide bridge between Cys-198 and Cys-206 via cal pH. Cys-51 belongs to the loop connecting ␤1 and ␣1. This the Cys-206 attack on Cys-198, previously implicated in the position at the entrance of ␣1 subjects the catalytic cysteine disulphide bond with Cys-51. The distance between their C␣ to the influence of the overall ␣ helix dipole and probably atoms is 20 A˚ in molecule A. The extended conformation of participates in its pKa decrease. Such a mechanism of activat- the C-terminal coil observed in this structure, if significant with ing a cysteine residue is often observed in proteins [37]. How- respect to the biologically active form, would require a large ever, other residues of the catalytic site could also contribute conformational change. Again, the flexibility can easily be en- to influence the Cys-51 pKa. visaged, especially since both cysteine residues were found The reduction of the sulphenic acid intermediate involves an to bracket a glycine-rich sequence [19]. All these observations Structure 1176

Figure 9. Location of the Cysteine Residues in the Structure of the E. coli PMSR The MsrA backbone of molecule A and of the four cysteine residues, Cys-51, Cys-86, Cys- 198, and Cys-206, displayed as ball-and-stick representations, are shown in stereo with MOLSCRIPT [48] and Raster3D [49].

are consistent with the existence of movements of the C-termi- tial to the catalytic mechanism that involves the nucleophilic nal tail and are compatible with the involvement of Cys-198 attack of a cysteine residue followed by two successive intra- and Cys-206 in the catalytic mechanism. molecular attacks by two other cysteine residues. Here we describe the first three-dimensional structure of a Biological Implications member of the PMSR family. The E. coli enzyme possesses 211 amino acids folded in a unique ␣/␤ motif plus a large Peptide methionine sulphoxide reductases (PMSRs) catalyze proportion of coil structures. The active site is a wide and open the reduction of methionine residues oxidized to sulphoxides basin covered with a large number of aromatic amino acids. by natural oxidative reagents. Depending on the localization The catalytic cysteine residue Cys-51, here observed in a caco- of the methionine residue in the protein structure, oxidation dylated form, probably has a high accessibility in its reduced of this amino acid leads to various effects. They range from form. The influence of the ␣ helix dipole on the Cys-51 pKa abolishment of the enzymatic activity to its increase and in- value might favor the presence of the thiolate form needed clude cases where no variation is observed. Whatever the pro- for the nucleophilic attack on the methionine sulphoxide, at cess involving this oxidation, sulphoxide reduction is essential physiological pH. The proton transfer involved in the release to restore the initial state, the methionine, in order to avoid of a water molecule from the sulphenic acid intermediate might cellular damages. Numerous organisms possess a PMSR, ei- occur via Glu-94 or Asp-129, which are situated beside Cys- ther as a unique domain or multidomain protein. Sequence 51. With Tyr-82 and Tyr-134, these amino acids interact with comparisons revealed several amino acids that are strictly con- two water molecules. One of them could mimic the position served across species. Some of them were shown to be essen- of the released water molecule. The next steps of the reaction

Table 2. Data Collection Statistics [SeMet]-MsrA MsrA Inflection Peak Remote Data Collection Statistics Wavelength (A˚ ) 0.9797 0.9794 0.9792 0.9724 Resolution (A˚ ) 30.0–1.9 24.7–3.0 24.7–3.0 24.7–3.0 Outer resolution shell (A˚ ) 1.97–1.90 3.16–3.00 3.16–3.00 3.16–3.00 Number observations 291,810 132,762 136,208 137,140 Number unique reflections 64,402 36,267 36,322 36,414 Completeness (%): overall (outer shell) 88.9 (65.6) 98.1 (93.5) 98.2 (94.2) 98.4 (96.0) a Rsym (%): overall (outer shell) 3.5 (23.5) 3.4 (6.7) 3.4 (6.9) 3.8 (6.5) Phasing Statistics Isomorphous phasing powerb acentric/centric 0.95/0.63 3.19/2.47 c Isomorphous RCullis acentric/centric 0.83/0.85 0.45/0.38 c Anomalous RCullis 0.40 0.45 Figure of merit (MLPHARE) 0.789 Figure of merit (DM) 0.888 a Rsym ϭ⌺⌺|Ii Ϫ Im|/⌺⌺Ii, where Ii is the intensity of the measured reflection and Im is the mean intensity of this reflection. b Phasing power ϭ FH/LOC, where LOC is the lack of closure, i.e. ||FP ϩ FHcalc| Ϫ |FPH||. c The Cullis R factor is defined here as RCullis ϭ LOC/⌺|FPH Ϯ FP|;FP,FPH and FH are the protein, derivative, and heavy-atom structure factors, respectively. MsrA and [SeMet]-MsrA data were respectively collected on beamlines BM30 and BM14 at the ESRF. Data were integrated and processed with DENZO and SCALEPACK [46]. Phase calculations were performed with MLPHARE [38]. Structure of Methionine Sulphoxide Reductase 1177

involve the formation of disulphide bridges between Cys-51 Schizosaccharomyces pombe (S, Q09859); Streptococcus gordonii (G, and Cys-198 and, after that, between Cys-198 and Cys-206. AAF36477); Streptococcus pneumoniae (S, P35593); Streptococcus pyo- Despite the distance between these cysteine residues, the (W, RST00032); Synechocystis sp. (P, S74472 and S74763); Trepo- nema pallidum (P, G71300); Ureaplasma urealyticum (G, AAF30698); Yersinia observed C-terminal flexibility is consistent with this hy- pestis (W, RYP00457). The multiple alignment was generated using CLUS- pothesis. TALW [45].

Experimental Procedures Acknowledgments

Crystallization and Data Collection We are very grateful to G. A. Leonard at the Joint Structural Biology Group MsrA and the fully selenomethionine-substituted MsrA ([SeMet]-MsrA) have at the ESRF for his important contribution to the MAD experiment. We been crystallized as previously described [36]. Crystals were grown in the warmly acknowledge Dr. Manfred Weiss at the Institute of Molecular Bio- presence of PEG8000, ammonium sulphate, cacodylate buffer, and DTT. technology in Jena (Germany) for the numerous discussions he had with ϭ ϭ ˚ They belong to space group P6522 with unit cell parameters a b 102.5 A, us. We thank the staffs of the beamlines BM14 and BM30 (ESRF, Grenoble) c ϭ 292.3 A˚ , ␥ϭ120Њ and have three molecules in the asymmetric unit that for their kind assistance during data collections. This research was sup- are related by no noncrystallographic symmetry n-fold axis. A native data ported by the Centre National de la Recherche Scientifique, the University set was collected on beamline BM30 at the ESRF to 1.9 A˚ resolution, and Henri Poincare´ Nancy I, the FR42 Prote´ ines, and the Association de la MAD measurements were performed to 3 A˚ resolution on beamline BM14 Recherche contre le Cancer (contract number 5436). at the ESRF at three different wavelengths corresponding to the inflexion point, peak, and high-energy remote of the selenium K-edge [36] (Table 2). Received: July 27, 2000 Revised: September 15, 2000 Structure Resolution and Refinement Accepted: September 22, 2000 Anomalous and dispersive differences calculated from the three MAD data sets were analyzed by CNS_SOLVE [23]. Since the number of molecules References per asymmetric unit was unknown at this stage of the structure determination and could range from 2 to 5, we searched for the position of a maximum 1. Vogt, W. (1995). Oxidation of methionyl residues in proteins: tools, tar- of 20 selenium atoms. Their position, occupancy, and B factor refinement, gets, and reversal. Free Radic. Biol. Med. 18, 93–105. for which we used MLPHARE [38] and the inflexion point data set as the 2. Brot, N., and Weissbach, H. (1983). Biochemistry and physiological role native, led us to use 18 of these sites for phase calculation. Later inspection of residues in proteins. Arch. Biochem. Biophys. of the final model revealed the correct assignment of the whole set of 223, 271–281. selenium atoms; i.e., their positions precisely corresponded to the positions 3. Brot, N., Weissbach, L., Werth, J., and Weissbach, H. (1981). Enzymatic refined for the 18 sulphur atoms of the 6 methionine residues contained in reduction of protein-bound methionine sulfoxide. Proc. Natl. Acad. Sci. the 3 molecules of the asymmetric unit. Phases were optimized by solvent USA 78, 2155–2158. flattening, with a 30% solvent mask automatically calculated by DM [39] 4. Rahman, M.A., Nelson, H., Weissbach, H., and Brot, N. (1992). Cloning, (Table 2). The high quality of the derived electron density map allowed full sequencing, and expression of the Escherichia coli peptide methionine chain tracing of one monomer and almost complete tracing for both others, sulfoxide reductase gene. J. Biol. Chem. 267, 15549–1551. for which we used O [40] and TURBO-FRODO [41]. This preliminary MsrA 5. Abrams, W.R., Weinbaum, G., Weissbach, L., Weissbach, H., and Brot, ˚ model was first refined to 3 A resolution. The selenium atoms were replaced N. (1981). Enzymatic reduction of oxidized ␣-1-proteinase inhibitor re- by sulphur atoms, and the model was then refined with the native data set stores biological activity. Proc. Natl. Acad. Sci. USA 78, 7483–7486. ˚ ˚ successively between 30.0 and 2.5 A, then between 30.0 and 1.9 A resolution. 6. Carp, H., et al., and Brot, N. (1983). Human methionine sulfoxide-peptide For this we used positional refinement and simulated annealing as imple- reductase, an enzyme capable of reactivating oxidized alpha-1-protein- mented in CNS_SOLVE [23]. In particular, bulk solvent refinement was per- ase inhibitor in vivo. Am. Rev. Respir. Dis 127, 301–305. ϭ ϭ ˚ 2 formed (refined values: K 0.35; B 36.2 A ). Isotropic B factors assigned 7. Sigalov, A.B., and Stern, L.J. (1998). Enzymatic repair of oxidative dam- to individual atoms were refined during the later stages of refinement. No age to human apolipoprotein A-I. FEBS Letters 433, 196–200. noncrystallographic symmetry restraints were used at any step of the refine- 8. Moskovitz, J., et al., and Weissbach, H. (1995). Escherichia coli peptide ment procedure. Water molecules were added in several steps with the methionine sulfoxide reductase gene: regulation of expression and role automated procedure implemented in CNS_SOLVE [23] followed by manual in protecting against oxidative damage. J. Bacteriol. 177, 502–507. control using O [40]. Statistics for the final model are shown in Table 1. 9. Moskovitz, J., Berlett, B.S., Poston, J.M., and Stadtman, E.R. (1997). The yeast peptide-methionine sulfoxide reductase functions as an anti- Sequence Alignment oxidant in vivo. Proc. Natl. Acad. Sci. USA 94, 9585–9589. Forty-one sequences of active or putative PMSRs were selected from the 10. Gabbita, S.P., Aksenov, M.Y., Lovell, M.A., and Markesbery, W.R. (1999). Swissprot [42], the Protein Information Resource (PIR) [43], or the GenPept Decrease in peptide methionine sulfoxide reductase in Alzheimer’s dis- [44] databases, or they were retrieved from the WIT website (http://wit.mcs. ease brain. J. Neurochem. 73, 1660–1666. anl.gov/WIT2/) developed in the Computational Biology Group at the Math- 11. Levine, R.L., Berlett, B.S., Moskovitz, J., Mosoni, L., and Stadtman, E.R. ematics and Computer Science Division at Argonne National Laboratory, (1999). Methionine residues may protect proteins from critical oxidative Argonne, IL. The species and, in parentheses, the corresponding accession damage. Mech. Ageing Dev. 107, 323–332. numbers and databases, referred to by their initial letters S, P, G, or W, are: 12. El Hassouni, M., Chambost, J.P., Expert, D., Van Gijsegem, F., and Arabidopsis thaliana (P, T05534); Bacillus subtilis (S, P54154); Bos taurus (S, Barras, F. (1999). The minimal gene set member msrA is a virulence P54149); Brassica napus (S, P54151); Caernorhabditis elegans (P, T34041); determinant of the plant pathogen Erwinia chrysanthemi. Biochemistry Campylobacter jejuni (G, CAB75273); Clostridium acetobutylicum (W, 96, 887–892. RCA03080); Deinococcus radiodurans (P, E75345); Drosophila melanogaster 13. Vogt, W., Zimmermann, B., Hesse, D., and Nolte, R. (1992). Activation (P, P08761); Enterococcus faecalis (W, REF02840); Erwinia chrysanthemi of the fifth component of human complement C5, without cleavage, by (G, CAA10143); E. coli (S, P27110); Fragaria ananassa (S, P54152); Gracilaria methionine oxidizing agents. Mol. Immunol. 29, 251–256. gracilis (G, AAD43253); Haemophilus influenzae (S, P45213); Helicobacter 14. Sun, H., Gao, J., Ferrington, D.A., Biesiada, H., Williams, T.D., and Squier, pylori (S, O25011); Homo sapiens (G, CAB59628); Lactuca sativa (G, T.C. (1999). Repair of oxidized calmodulin by methionine sulfoxide re- AAF19789); Lycopersicon esculentum (S, P54153); Methanobacterium ductase restores ability to activate the plasma membrane Ca-ATPase. thermoautotrophicum (P, F69170); Mycobacterium leprae (W, RML01550); Biochemistry 38, 105–112. Mycobacterium tuberculosis (P, D70616); Mycoplasma capricolum (P, 15. Ciorba, M.A., Heinemann, S.H., Weissbach, H., Brot, N., and Hoshi, T. S77828); Mycoplasma genitalium (S, P47648); Mycoplasma pneumoniae (S, (1997). Modulation of potassium channel function by methionine oxida- P75188); Neisseria gonorrhoeae (W, RNG01851); Neisseria meningitidis (G, tion and reduction. Proc. Natl. Acad. Sci. USA 94, 9932–9937. CAB83597); Porphyromonas gingivalis (W, RPG01276); Pseudomonas aeru- 16. Wizemann, T.M., et al., and Masure, H.R. (1996). Peptide methionine ginosa (W, RPA04229); Pseudomonas fluorescens (G, AAC15512); Rhodo- sulfoxide reductase contributes to the maintenance of adhesins in three bacter capsulatus (W, RRC00408); Saccharomyces cerevisiae (S, P40029); major pathogens. Proc. Natl. Acad. Sci. USA 93, 7985–7990. Structure 1178

17. Montgomery, J., Goldman, S., Deikman, J., Margossian, L., and Fischer, methods for building protein models in electron density maps and for R.L. (1993). Identification of an ethylene-responsive region in the pro- the location of errors in these models. Acta Crystallogr. A 47, 110–119. moter of a fruit ripening gene. Proc. Natl. Acad. Sci. USA 90, 5939–5943. 41. Roussel, A., and Cambillau, C. (1992). The TURBO-FRODO Graphics 18. Moskovitz, J., Poston, J.M., Berlett, B.S., Nosworthy, N.J., Szczepanow- Package. In Silicon Graphics Geometry Partners Directory, vol. 81, Sili- ski, R., and Stadtman, E.R. (2000). Identification and characterization con Graphics, ed. (Mountain View, CA: Silicon Graphics). of a putative active site for peptide methionine sulfoxide reductase 42. Bairoch, A., and Apweiler, R. (2000). The SWISS-PROT protein sequence (MsrA) and its substrate stereospecificity. J. Biol. Chem. 275, 14167– database and its supplement TrEMBL in 2000. Nucleic Acids Res. 28, 14172. 45–48. 19. Lowther, W.T., Brot, N., Weissbach, H., Honeck, J.F., and Matthews, 43. Barker, W.C., et al., and Wu, C. (2000). The Protein Information Resource B.W. (2000). Thiol-disulphide exchange is involved in the catalytic mech- (PIR). Nucleic Acids Res. 28, 41–44. anism of peptide methionine sulfoxide reductase. Proc. Natl. Acad. Sci. 44. Benson, D.A., et al., and Wheeler, D.L. (1999). GenBank. Nucleic Acids USA 97, 6463–6468. Res. 27, 12–17. 20. Boschi-Muller, S., Azza, S., Sanglier-Cianferani, S., Talfournier, F., Van 45. Thompson, J.D., Higgins, D.G., and Gibson, T.J. (1994). CLUSTALW: Dorsselear, A., and Branlant, G. (2000). A sulfenic acid enzyme interme- improving the sensitivity of progressive multiple sequence alignment diate is involved in the catalytic mechanism of Peptide Methionine Sulf- through sequence weighting, position-specific gap penalties and weight oxide Reductase from Escherichia coli. J. Biol. Chem., in press. matrix choice. Nucleic Acids Res. 22, 4673–4680. 21. Bru¨ nger, A.T. (1992). The free R value: a novel statistical quantity for 46. Otwinowski, Z., and Minor, W. (1997). Processing of X-ray diffraction assessing the accuracy of crystal structures. Nature 355, 472–475. data collected in oscillation mode. In Methods in Enzymology, vol. 276, 22. Laskowski, R.A., MacArthur, M.W., Moss, D.S., and Thornton, J.M. C.W. Carter, Jr. and R.M. Sweet, eds. (New York: Academic Press), pp. (1993). PROCHECK: a program to check the stereochemical quality of 307–326. protein structures. J. Appl. Crystallogr. 26, 282–291. 47. Westhead, D.R., Slidel, T.W., Flores, T.P., and Thornton, J.M. (1999). 23. Bru¨ nger, A.T., et al., and Warren, G.L. (1998). Crystallography & N.M.R. Protein structural topology: automated analysis and diagrammatic rep- system: a new software suite for macromolecular structure determina- resentation. Protein Sci. 8, 897–904. tion. Acta Crystallogr. D 54, 905–921. 48. Kraulis, P.J. (1991). MOLSCRIPT: a program to produce both detailed 24. Ramakrishnan, C., and Ramachandran, G.N. (1965). Stereochemical cri- and schematic plots of protein structures. J. Appl. Crystallogr. 26, teria for polypeptide and protein chain conformation. Biophys. J. 5, 282–291. 909–933. 49. Meritt, E.A., and Murphy, M.E.P. (1994). Raster3D version 2.4 – a pro- 25. Zhang, C.-T., and Zhang, R. (1999). Skewed distribution of protein sec- gram for photorealistic molecular graphics. Acta Crystallogr. D 50, ondary structure contents over the conformational triangle. Protein En- 869–873. gineering 12, 807–809. 50. Gouet, P., Courcelle, E., Stuart, D.I., and Metoz, F. (1999). ESPript: 26. Cuff, J.A., Clamp, M.E., Siddiqui, A.S., Finlay, M., and Barton, G.J. (1998). multiple sequence alignments in PostScript. Bioinformatics 15, 305–308. Jpred: a Consensus Secondary Structure Prediction Server. Bioinfor- matics 14, 892–893. Accession Numbers 27. Berman, H.M., et al., and Bourne, P.E. (2000). The Protein Data Bank. Nucleic Acids Res. 28, 235–242. The atomic coordinates have been deposited with the Brookhaven Protein 28. Shindyalov, I.N., and Bourne, P.E. (1998). Protein structure alignment Data Bank [27], accession code 1FF3. by incremental combinatorial extension (CE) of the optimal path. Protein Engineering 11, 739–747. 29. Kohls, D., Sulea, T., Purisima, E.O., MacKenzie, R.E., and Vrielink, A. (2000). The crystal structure of the formiminotransferase-cyclodeami- nase: implications for substrate channeling in a bifunctional enzyme. Structure Fold Des. 8, 35–46. 30. Ermler, U., Merckel, M.C., Thauer, R.K., and Shima, S. (1997). Formyl- methanofuran: tetrahydromethanopterin formyltranferase from Metha- nopyrus kandleri – new insights into salt-dependence and thermostabil- ity. Structure Fold Des. 5, 635–646. 31. Moskovitz, J., et al., and Brot, N. (1996). Chromosomal localization of the mammalian peptide-methionine sulfoxide reductase gene and its differential expression in various tissues. Proc. Natl. Acad. Sci. USA 93, 3205–3208. 32. Sadanandom, A., et al., and Fairbairn, D.J. (1996). Identification of a peptide methionine sulphoxide reductase gene in an oleosin promoter from Brassica napus. Plant J. 10, 235–242. 33. Kuschel, L., et al., and Heinemann, S.H. (1999). Molecular cloning and functional expression of a human peptide methionine sulfoxide reduc- tase (hMsrA). FEBS Lett. 456, 17–21. 34. Taha, M.K., So, M., Seifert, H.S., Billyard, E., and Marchal, C. (1988). Pilin expression in Neisseria gonorrhoeae is under both positive and negative transcriptional control. EMBO J. 7, 4367–4378. 35. Maignan, S., Guilloteau, J.P., Zhou-Liu, Q., Cle´ ment-Mella, C., and Mikol, V. (1998). Crystal structures of the catalytic domain of HIV-1 integrase free and complexed with its metal cofactor: high level of similarity of the active site with other viral integrases. J. Mol. Biol. 282, 359–368. 36. Teˆ te-Favier, F., et al., and Aubry, A. (2000). Crystallisation and prelimi- nary X-ray diffraction studies of the peptide methionine sulfoxide reduc- tase from Escherichia coli. Acta Crystallogr. D 56, 1194–1197. 37. Hol, W.G.J. (1985). The role of the ␣ helix dipole in protein function and structure. Prog. Biophys. Mol. Biol. 45, 149–195. 38. CCP4 (Collaborative Computational Project 4) (1994). The CCP4 suite: programs for protein crystallography. Acta Crystallogr. D 50, 760–763. 39. Cowtan, K. (1994). DM: an automated procedure for phase improvement by density modification. Joint CCP4 and ESF-EACBM Newsletter on Protein Crystallography 31, 34–38. 40. Jones, T.A., Zou, J.Y., Cowan, S.W., and Kjeldgaard, M. (1991). Improved